notebooklm-mcp-programmatic-access

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow includes adding sources from arbitrary URLs (nlm source add --url ... / MCP source_add), which causes the system to fetch and ingest outsider-authored web page text into NotebookLM (public web content fetched at runtime → LLM context via NotebookLM/MCP query results).

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill exposes runtime commands that fetch arbitrary web content into NotebookLM (e.g., "nlm source add --url https://news.ycombinator.com" and "https://arxiv.org/list/cs.AI/recent"), and that fetched content is injected as notebook context used to answer queries, so external URLs can directly control agent prompts.